The environmental impacts of data centers
Dr. Enid Sisskin discusses the extensive environmental impacts of AI data centers, including significant water and electricity consumption, land use, noise, and electronic waste. She highlights methods to mitigate these harms and notes that some areas, like Escambia and Santa Rosa counties, have enacted bans on data centers. The article also mentions Florida's new state law, SB 484, regulating hyperscale data centers.
Dr. Enid Sisskin, during an EcoMinutes podcast, addressed the environmental impacts of Artificial Intelligence and the large data centers required to support it, a topic suggested by a listener. While AI offers environmental benefits such as enhancing energy efficiency and monitoring, its infrastructure carries significant direct and indirect environmental costs.
Data centers are major consumers of water, particularly for cooling, with US data centers projected to use 8% of public water by 2030, largely from potable sources. They also demand massive amounts of electricity, with hyperscale data centers using as much power as 100,000 homes, exemplified by Meta's planned facilities in Louisiana and Wyoming. This surging demand outpaces grid capacity and renewable energy growth, delaying fossil fuel plant retirement and contributing to climate change. Additionally, data centers require extensive land, contribute to noise pollution (up to 100 dB from outdoor equipment), and generate substantial electronic waste containing hazardous substances.
Sisskin emphasized that many environmental impacts can be mitigated, advocating for sustainable construction, circularity practices, water neutrality through advanced cooling and reuse strategies, and on-site renewable energy. She also suggested financing grid improvements and implementing noise reduction measures. She noted a growing bipartisan movement to ban data centers, citing current bans in Escambia and Santa Rosa counties and Florida's new law, SB 484, which regulates hyperscale and large-scale data centers.
Ultimately, Sisskin encouraged listeners to pressure local governments to insist on mitigation efforts and to reduce their own AI usage, highlighting the significant energy and water consumption of AI queries and email generation.